As part of this year’s Sustainability Week in Lund, the Lund Stem Cell Centre at Lund University’s Faculty of Medicine invited both the public and the research community to explore the responsible use of AI in life science research in the panel discussion “Sustainable AI in Practice: From Computational Costs to Real-World Impacts”.
Bringing together expertise in machine learning, high-performance computing and AI hardware, the discussion explored the sources of AI’s environmental impact and how AI can support more sustainable research.
Daniel Twohig and Claire McKay, two of the organisers and members of the Lund Stem Cell Centre’s Sustainability Committee, highlighted why this conversation was timely.
Why is sustainability an important issue when we talk about artificial intelligence in research today?
“While AI is profoundly transforming our world by accelerating scientific discovery and data handling, as well as potentially boosting productivity, sustainability is certainly a critical issue as AI becomes increasingly prevalent in research and other industries,” explains Claire McKay.
“The environmental footprint, in terms of carbon emissions, water use for cooling data centres and electronic waste, is becoming a major challenge. AI is a powerful and useful research tool, but it can also divert budget, attention and resources away from achieving sustainable development.”
Many researchers may not consider the computational or energy costs of AI tools. What do you hope this event would make people more aware of?
“We hope that this event inspired the responsible use of AI and encourage people to avoid frivolous applications and consider alternative approaches, such as downsizing to more efficient compressed or small language models,” says Claire McKay.
Beyond environmental costs, what other real-world impacts of AI did the panel discussion highlight?
The panel also addressed broader societal effects of AI, including shifts in employment trends and potential impacts on human health and capacity.
“Employment trends might be a particularly important issue,” Claire McKay notes.
Daniela Twohig added: “One of academia’s primary roles is to prepare people for various careers. Thus, universities may need to overhaul a multitude of courses and even some departments if AI predictions hold true.”
What is one key takeaway you hope attendees brought back to their own research or studies?
“We hope that attendees left with a clearer and more informed understanding of what is happening in academia and industry, from both local and international perspectives,” says Daniel Twohig.
“The differing vantage points of our panellists and moderator aimed to inspire people to begin, or improve upon, their efforts to actively work towards employing more sustainable computation in their own research. Moreover, it would be great if some attendees felt encouraged to dive deeper into the field."